The entire corpus, in one search.
Including papers published hours ago. Stay current, build reading lists, verify claims, and map collaborators, all in the assistant you already use.
See it work on real questions →
Current research, inside the tools you already use.
Valency Bond™ connects AI assistants and research workflows to the scientific literature, so researchers can find, verify, and cite trusted work without changing how they work.
Get free accessBuilt for researchers first
Bond serves three audiences, in this order: the researchers doing the work, the institutions that support them, and the builders wiring AI into science. Everyone draws on the same grounded corpus.
Internal research, connected globally.
Your private research, connected to global scientific knowledge, without any of it leaving your secure environment. Give every team one grounded view of the literature, and give your agents a source of truth they can cite.
Talk to usBond Valency to any LLM.
Pull live scientific knowledge into Claude, Cursor, or any MCP-compatible agent. Responses are token-efficient and blazingly fast, so grounding your model spends context, not your whole budget.
Discover. Understand. Analyze. Deliver.
The same 38+ tools that ground a single chat answer also power a full research workflow, from first search to final export.
Discover
Find work by meaning or metadata across the full corpus. Semantic search reaches past keyword matching to the ideas underneath, so the right paper surfaces even when it uses different words than you did.
Understand
Profile researchers, map co-authors, pull full paper histories. Follow the citation graph in both directions to see what a result built on and what has built on it since.
Analyze
Track publication and keyword momentum in any field. See what is heating up, what is fading, and who is driving it, before it shows up in a review article.
Deliver
Filter precisely, then export clean JSON or CSV, ready to cite. Hand your model a tidy, structured result set, or drop it straight into your reference manager.
What the artery delivers
Three things a reasoning model cannot get on its own: work that is fresh, sources that are real, and a connection that fits whatever tool you already run.
Fresh within hours
Updated every few hours with the latest published research. Search work published hours ago, not months.
Grounded in real research
Connected to trusted data sources, including arXiv, bioRxiv, medRxiv, PubMed, PsyArXiv, SocArXiv, EarthArXiv, and more. Grounded in reality, not hallucinations.
Any LLM, any workflow
Whether you're using Gemini, ChatGPT, or Claude, our connector plugs into your scientific and research workflows.
Papers in, grounded answers out
Bond is the middle box. Trusted papers and preprints flow in, get structured and indexed, and flow back out as grounded, citable context for your models and agents.
Papers & preprints
Fresh from trusted sources
Valency Bond™
Structured, indexed, grounded
LLMs & agents
Answers with citations
Connect in three steps
No SDK to learn and no pipeline to build. If your assistant speaks MCP, you are minutes away.
Sign up
Create a Valency account. Researchers, academics, and non-profits use Bond for free.
Connect your LLM
Point Claude, ChatGPT, Kimi Code, or any MCP-friendly LLM at Bond. Auth is OAuth 2.0 with PKCE, and setup takes under three minutes.
Ask, grounded
Ask your questions the way you always have. Now the answers come back with real, current sources attached, ready to cite.
What it costs
Researchers, academics, and non-profits
The full corpus and all 38+ tools, at no cost. Sign up and connect your assistant.
Get free accessInstitutions and R&D teams
Internal research connected to the global literature, without any of it leaving your secure environment.
Talk to us
Built to plug in
Bond is an open, standards-based connector, not a walled garden. If your tools speak MCP, they already speak Bond.
One connector, 38+ tools
A single endpoint exposes search, researcher profiles, citation graphs, trends, and export, so you add one integration instead of ten.
Connect in minutes
Connected in under 3 minutes. Authenticate once, point your assistant at Bond, and start asking grounded questions.
Standards-based
Speaks MCP. OAuth 2.0 with PKCE for auth, and clean JSON-LD responses your tools and models can parse without glue code.
Bond, in brief
The questions researchers and teams ask us most.
Is Bond a new AI model?
No. Bond is the grounded data layer behind the models you already use. It does not try to out-reason your LLM. It makes sure your LLM is reasoning from the real, current literature.
Which assistants and agents does it work with?
Anything that speaks the model context protocol, including Claude, ChatGPT, and Gemini, plus custom agents in your own harness. One connector, any LLM.
What does it cost?
Bond is free for researchers, academics, and non-profits. Institutions and R&D teams can talk to us about private deployments.
Where does the data come from, and how fresh is it?
Trusted sources including arXiv, bioRxiv, medRxiv, PubMed, PsyArXiv, SocArXiv, EarthArXiv, and more. The corpus is updated every few hours, so new work is reachable within hours of publication.
Will it fit in my token budget?
Yes. Responses are token-efficient and structured, so grounding a model adds signal without flooding its context window.
Is my private research exposed?
No. Institutions can connect internal research to global knowledge without any of it leaving their secure environment.